Skip to main content

Learn About Einstein Classification Apps

Learning Objectives

After completing this unit, you’ll be able to:

  • Describe predictive intelligence.
  • Identify the benefits of Einstein classification apps.
  • Begin planning for case classification and case wrap-up.

Ursa Major Solar’s business is booming. Solar panel sales have skyrocketed. The business has expanded beyond the Southwest, which means keeping thousands more customers happy and loyal. The CEO of Ursa Major Solar, Sita Nagappan-Alvarez, spotted a statistic from Salesforce Research that intrigued her: 84% of customers say that the experience a company provides is as important as its products and services.


Sita uses Service Cloud to help her company create exceptional customer experiences. The service platform:

  • Unifies case management across many engagement channels, such as email, phone, and the web.
  • Helps customers and support agents access, capture, and share useful information with a knowledge base.
  • Uses bots and artificial intelligence (AI) to quickly answer customers’ frequently asked questions online.

Sita wonders what else she can do to improve service. She asks her rock-star admin, Maria Jimenez, for advice.

Sita and Maria standing alongside the Ursa Major Solar logo.

Maria explains that according to Salesforce Research, most customers are open to companies using AI to improve the experience they receive. AI makes your daily experiences smarter by embedding predictive intelligence into your everyday apps. 

What Is Einstein Classification?

Enter the Einstein classification apps—Einstein Case Classification and Einstein Case Wrap-Up. These apps help your agents at both ends of the case lifecycle: when a case is created and when a chat with a customer ends. They apply predictive intelligence to “classify” or suggest case field values based on values from your closed cases. Using machine learning—AI technology that improves prediction accuracy over time—classification reduces guesswork in completing case fields, and frees up time for your support agents. Less time spent on paperwork means more time dedicated to your customers.

You can set up and use one or both apps. Einstein Case Classification provides recommendations once, right after a case is created. Einstein Case Wrap-Up, shows agents field predictions on demand or when a chat conversation ends. 


Einstein's recommendations appear in the Case Classification component (1) or the Case Wrap-Up Einstein Field Recommendations component (2). 

Einstein’s recommendations in a highlighted Einstein Field Recommendations components.

You have full control of how field predictions are applied. For your preferred fields, Einstein recommends the top three values for agents to choose from, or you can have Einstein select the best value for an agent to confirm. For Einstein Case Classification, you can set a threshold where Einstein selects and saves field values automatically. And after classification, Einstein can run case assignment rules on auto-updated cases so that they’re routed to the right agent.    

To control when Einstein selects or saves field values, set thresholds for prediction confidence. The more automated the action, the higher the prediction confidence needed. And if a prediction doesn’t meet one option’s threshold, Einstein tries the next, less automated approach. At the highest confidence levels, field values are automatically filled. At lower confidence levels, Einstein preselects options for the agent to confirm, or presents the agent with three recommendations to choose from. 

Einstein Classification and Artificial Intelligence

Because these classification apps are based on machine learning, Einstein's predictions are only as good as the data it’s given. Most machine learning relies on human beings to identify and describe features in a data set—in this case, the closed cases in your Salesforce org. The accuracy of data in your closed cases, and the number of closed cases used to build the learning model, help Einstein predict field values correctly. 


A typical machine learning solution can have thousands or even millions of hand-designed features. After humans complete the initial identification work by hand, the machine uses a learning algorithm to adjust the weighting of each feature to make more accurate predictions.  

Benefits of Einstein Classification Apps

As CEO, Sita sees that Einstein classification apps support her vision of delivering exceptional service experiences to customers, leading to more sales and bolstering customer loyalty.

  Benefit   Description
Saved time for agents As support agents work on cases, they spend less time scrolling and searching for the right field values. Automatic case routing lets agents focus on high-order tasks.
Improved data quality The predictive model improves data accuracy on cases because there’s less likelihood for human error.
Faster case resolution Since cases are automatically classified based on user histories and trends, cases can route to the right support agents for quicker resolution.
Better customer service More time for agents, improved data accuracy, and faster case resolution lead to more focus on building strong customer relationships and increasing customer satisfaction (CSAT) scores.

Plan for Classification

Adding AI is the fourth stage of the general setup process for Service Cloud. (See the Service Cloud for Lightning Experience module for a refresher.)

Service Cloud’s implementation process represented by 4 steps: case management, channels, knowledge, and AI and bots.

Maria starts by setting up case management features for Ursa Major Solar before implementing AI. Einstein is smart and can help you classify your cases, but it can't define the business process for managing cases at Ursa Major Solar; Maria has to do that. She starts by determining how to capture the right information and who should work on each case, setting required fields, notification preferences, assignment rules, and routing. 


Although Maria knows that Ursa Major Solar is ready for AI, before she clicks anything in Setup, she meets with the service team to learn some details about how they operate.

  Question   Answer
Which picklist, checkbox, or lookup fields on cases are best suited to predictive intelligence? Some useful fields to predict are Case Reason, Language, Escalated, and Priority. Einstein can also predict the values of custom picklist, checkbox, and lookup fields.

To predict Language, you don't need to have a historic case data set in that language.
For each field that you want to predict for your agents, are there at least 400 closed cases with a value in that field? Hmm… Someone will have to look into that. Yes, we more than likely have 400 closed cases that use each field we want to predict.
Who on the team can review closed case data to ensure its accuracy before we use it to build a predictive model? Maybe Ryan De Lyon? He’s a customer service manager and knows about cases.
Do we want Einstein to automatically save field values, or should support agents review recommendations first? For now, let’s have our agents review recommendations first. Later, let’s consider automating field values that Einstein predicts with high confidence.
Have we identified specific support agents who should have access to case classification? Yes. Our Tier 1 support team should have access, so we can assign them the Einstein Case Classification User permission set during implementation.


With this bit of planning done, Maria is ready to take the next step to implement case classification for Ursa Major Solar—preparing data to build a predictive model.

Resources

Rights of ALBERT EINSTEIN are used with permission of The Hebrew University of Jerusalem. Represented exclusively by Greenlight.

Keep learning for
free!
Sign up for an account to continue.
What’s in it for you?
  • Get personalized recommendations for your career goals
  • Practice your skills with hands-on challenges and quizzes
  • Track and share your progress with employers
  • Connect to mentorship and career opportunities